?
Dimension-Augmented Anisotropy in Graph Neural Diffusion
Ch. 4. P. 29–33.
Sycheva T., Beketov M., Smolyar I.
We consider Graph Anisotropic Diffusion (GAD), a recently proposed model of graph neural networks, that can be trained to predict desired properties of the graph by performing learnable diffusion of node features on it. In contrast with similar methods, GAD introduces anisotropy of said diffusion by incorporating filters built from the graph’s Fiedler vector. In present work we attempt to improve this approach by increasing the dimension of the space in which GAD runs – that is, adding filters built from other low-frequency eigenmodes of the graph (eigenvectors of its Laplacian). We report the performance of such “dimension-augmented” GAD in predicting the chemical properties of small organic molecules from the ZINC dataset.
Keywords: graph Laplacianлапласиан графаdrug designграфовые нейронные сетиgraph neural networks (GNNs)дизайн лекарств
Publication based on the results of:
Solovev G. V., Zhidkovskaya A. B., Orlova A. et al., , in: Findings of the Association for Computational Linguistics: EMNLP 2025.: Association for Computational Linguistics, 2025. Ch. 367 P. 6956–6998.
Hit identification is a central challenge in early drug discovery, traditionally requiring substantial experimental resources. Recent advances in artificial intelligence, particularly large language models (LLMs), have enabled virtual screening methods that reduce costs and improve efficiency. However, the growing complexity of these tools has limited their accessibility to wet-lab researchers. Multi-agent systems offer a promising ...
Added: November 16, 2025
Parakal E. G., Kuznetsov S., Makarov I. et al., IEEE Access 2025 Vol. 13 P. 149657–149678
This paper proposes a novel explainable document classification framework that integrates Concept Whitening (CW) with graph concepts that are derived from stable graph patterns, and extracted via methods based on Formal Concept Analysis (FCA) and pattern structures. Document graphs are constructed using Abstract Meaning Representation (AMR) graphs, from which graph concepts are extracted and aligned ...
Added: October 22, 2025
Yusupov V., Rakhuba M., Frolov E., , in: CIKM '25: Proceedings of the 34rd ACM International Conference on Information and Knowledge Management.: ACM, 2025. Ch. 1 P. 5469–5473.
In this work, we present a fast and effective Linear approach for updating recommendations in a scalable graph-based recommender system UltraGCN. Solving this task is extremely important to maintain the relevance of the recommendations under the conditions of a large amount of new data and changing user preferences. To address this issue, we adapt the ...
Added: October 3, 2025
Perepelkin A., Sharifov A., Titov D. et al., Energies 2025 Vol. 18 No. 14 Article 3881
In the gas industry, accurate forecasting of gas production is critical for optimizing well operating conditions. Although traditional hydrodynamic models offer high accuracy, they are often computationally intensive and time-consuming, prompting a growing interest in proxy-based alternatives. This study proposes a hybrid methodology based on Spatio-Temporal Graph Neural Networks (ST-GNNs) for gas production forecasting. The ...
Added: July 23, 2025
Nikita Severin, Savchenko A., Kiselev D. et al., , in: RecSys '23: Proceedings of the 17th ACM Conference on Recommender Systems.: Association for Computing Machinery (ACM), 2023.
Recommender systems are essential for personalized content delivery and have become increasingly popular recently. However, traditional recommender systems are limited in their ability to capture complex relationships between users and items. Dynamic graph neural networks (DGNNs) have recently emerged as a promising solution for improving recommender systems by incorporating temporal and sequential information in dynamic ...
Added: May 22, 2025
Krivonosov M., Nazarenko T., Ushakov V. et al., Technologies 2025 Vol. 13 No. 1 Article 13
This paper introduces a novel approach for classifying multidimensional physiological and clinical data using Synolitic Graph Neural Networks (SGNNs). SGNNs are particularly good forto addressing the challenges posed by high-dimensional datasets, particularly in healthcare, where traditional mMachine lLearning and Artificial Intelligence methods often struggles to find global optima due to the “curse of dimensionality”. To ...
Added: December 23, 2024
Buikin Petr, Vologzhanina A., Novikov R. et al., Pharmaceutics 2023 Vol. 15 No. 9 Article 2180
Abiraterone acetate (AbirAc) is the most used steroidal therapeutic agent for treatment of prostate cancer. The mainly hydrophobic molecular surface of AbirAc results in its poor solubility and plays an important role for retention of abiraterone in the cavity of the receptor formed by peptide chains and heme fragments. In order to evaluate the hydrolytic ...
Added: May 24, 2024
D'yakonov A., Меведев А. В., Доклады Российской академии наук. Математика, информатика, процессы управления (ранее - Доклады Академии Наук. Математика) 2023 Т. 514 № 2 С. 118–125
For prediction tasks using tabular data, it is possible to extract additional information about the target variable by examining the relationships between the objects. Specifically, if it is possible to receive agraph
in which the objects are represented as vertices and the relationships are expressed as edges, then it is likely
that the graph structure contains valuable ...
Added: March 18, 2024
Khramchikhin A., Skryl’nikova M., Esaulkova I. et al., Molecules 2022 Vol. 27 No. 22 Article 7940
This study of the interaction system of binucleophilic 3-substituted 4-amino-4H-1,2,4-triazole-5-thiols and 3-phenyl-2-propynal made it possible to develop a new approach to synthesis of such isomeric classes as 7-benzylidene-[1,2,4]triazolo[3,4-b][1,3,4]thiadiazine and 8-phenyl-[1,2,4]triazolo[3,4-b][1,3,4]thiadiazepine. Among the 20 compounds studied in vitro against influenza A/Puerto Rico/8/34 (H1N1) virus, half of them demonstrated selectivity index (SI) of 10 or higher and ...
Added: February 8, 2024
Lukin R., Grigoriev R., Yarullin A. et al., , in: Proceedings of the NeurIPS 2021 Competitions and Demonstrations Track.: PMLR, 2022. P. 1–12.
In this report, we describe the Open Catalyst Challenge held at NeurIPS 2021, focusing on using machine learning (ML) to accelerate the search for low-cost catalysts that can drive reactions converting renewable energy to storable forms. Specifically, the challenge required participants to develop ML approaches for relaxed energy prediction, i.e. given atomic positions for an ...
Added: January 12, 2024
Oleg Platonov, Denis Kuznedelev, Michael Diskin et al., , in: Proceedings of the 11th International Conference on Learning Representations (ICLR 2023).: ICLR, 2023.
Added: December 28, 2023
Magaj G., Proceedings of Machine Learning Research 2023 Vol. 221 No. 3 P. 1–6
This paper presents the computational challenge on topological deep learning that was hosted within the ICML 2023 Workshop on Topology and Geometry in Machine Learning. The competition asked participants to provide open-source implementations of topological neural networks from the literature by contributing to the python packages TopoNetX (data processing) and TopoModelX (deep learning). The challenge ...
Added: October 26, 2023
Medvedev M., Stroganov O., Dmitrienko A. et al., Mendeleev Communications 2022 Vol. 32 No. 6 P. 735–738
Here we propose an over-the-hood docking method that compensates for systematic errors in the docking force fields. This method explicitly estimates the interaction energy of the ligand with the protein surface and uses it as a baseline to estimate the actual binding energy in the active site. It improves the accuracy of virtual screening in the LeadFinder package by up ...
Added: October 25, 2022
Pich i Rosello O., Vlasova A., Shichkova P. et al., / Series 005140 "Biorxiv". 2017.
Human genetic variability is thought to account for a substantial fraction of individual biochemical characteristics — in biomedical sense, of individual drug response. However, only a handful of human genetic variants have been linked to medication outcomes. Here, we combine data on drug-protein interactions and human genome sequences to assess the impact of human variation ...
Added: July 7, 2017
Tikhonov K., Mirlin A. D., Physical Review B: Condensed Matter and Materials Physics 2016 Vol. 94 P. 184203-1–184203-16
We investigate analytically and numerically eigenfunction statistics in a disordered system on a finite Bethe lattice (Cayley tree). We show that the wave-function amplitude at the root of a tree is distributed fractally in a large part of the delocalized phase. The fractal exponents are expressed in terms of the decay rate and the velocity ...
Added: March 9, 2017
Yurii Burman, Ploskonosov A., Trofimova A., Linear Algebra and its Applications 2015 No. 466C P. 64–82
We calculate characteristic polynomials of operators explicitly presented
as polynomials of rank 1 operators. Corollaries of the main result
(Theorem 1) include a generalization of the Forman's formula
for the determinant of the graph Laplacian, the
celebrated Matrix-tree theorem by G.Kirchhoff, and some
its generalizations and analogs, both known (e.g. the Matrix-hypertree
theorem by G.Masbaum and A.Vaintrob) and new. ...
Added: April 4, 2014
Burman Y. M., / Series math "arxiv.org". 2012. No. arXiv:1205.1123.
We calculate characteristic polynomials of operators explicitly represented as polynomials of rank $1$ operators. Applications of the results obtained include a generalization of the Forman--Kenyon's formula for a determinant of the graph Laplacian and also provide its level $2$ analog involving summation over triangulated nodal surfaces with boundary. ...
Added: May 15, 2012